This past September, Olivier Elemento, PhD, Associate Director of the Institute for Computational Biomedicine and Director of the Laboratory of Cancer Systems Biology at Weill Cornell Medicine in New York, was named Director of Weill Cornell’s Englander Institute for Precision Medicine. In this position he plans to work to expand the institution’s commitment to utilizing whole-exome and whole-genome sequencing to advance scientific knowledge and create personalized therapy for patients with cancer, especially those with advanced disease.

In 2016, Dr. Elemento and Mark A. Rubin, MD, then Founding Director of the Englander Institute for Precision Medicine (currently at the University of Bern, Switzerland) and Professor of Pathology and Laboratory Medicine at Weill Cornell Medicine, and their colleagues, published a clinical validation study of the Exome Cancer Test (EXaCT)-1. EXaCT-1 was developed by members of the Englander Institute team and won approval for its use from the New York State Department of Health in 2015.

The test, which involves sequencing both tumor and matched constitutional DNA to distinguish somatic mutations in the tumor from background germline mutations, can detect genomic alterations driving specific tumors. The data collected from -EXaCT-1 are used to determine the most likely effective therapy for patients.

In their validation study, EXaCT-1 was tested on 57 tumors covering 5 distinct clinically relevant mutations. A prospective analysis of 337 cancer cases found mutations in clinically relevant genes in 82% of the tumors, demonstrating the test is an accurate method for identifying actionable mutations.¹

The next generation of the whole-exome sequencing test, EXaCT-2, combines broad exome sequencing with targeted deeper sequencing of dozens of highly actionable genes, specific promoters, and viral sequences. EXaCT-2 is expected to be available for use in the clinic by mid-2018.

The ASCO Post talked with Dr. Elemento about his plans to expand Weill Cornell’s whole-exome and whole-genome sequencing assays to provide more patients with cancer—especially those with advanced-stage disease—better treatment options. Dr. Elemento also discussed what his research is showing about how to more accurately predict which patients are most likely to benefit from immunotherapy.

Precision Medicine in Advanced Cancers

Please talk about how you are planning to expand your institution’s research efforts in precision medicine for patients with advanced cancer.

We have had an ambitious precision medicine program for oncology care for the past few years. We have focused our efforts on whole-exome sequencing for patients with advanced cancers, because there is more opportunity for precision medicine to have an impact in these patients. These are patients who have few or no clinical options left, and we are looking to genomics to tell us how we can find additional treatment options for them.

Currently, we are limiting our screening to whole-exome sequencing. But very soon, we will be performing whole-genome sequencing on most patients with late-stage cancers.

Role in Early-Stage Disease

Will you also be using the technology to screen patients with early-stage disease?

We are focused right now on patients with metastatic cancers because they are in dire need of new therapeutic options. But we are very much looking to apply the precision medicine paradigm to patients with earlier-stage cancers to learn what we can do in addition to chemotherapy, surgery, or radiation therapy that would specifically target the mechanisms responsible for the cancer, and so we can individualize treatment and potentially induce longer remissions and perhaps even cures in more patients.

Newly diagnosed and early-stage patients have disease that is in theory relatively simpler to treat than advanced disease, which may have been treated with multiple drugs and has likely become highly heterogeneous. The reality is for patients with meta-static disease, even if you can identify an actionable mutation, they will be given drugs with toxicities and may be unable to tolerate those toxicities.

Interpreting Cancer Genomes

How are you using computational biology to advance precision medicine in patients with cancer?

Precision medicine relies completely on computational biology for the interpretation of genomic data and how to apply that information to individual patients. When we look at a patient’s cancer genome, we see a lot of mutations, but we are not always able to connect those mutations to a particular response to treatment.

We are using computational biology as a way to establish the connection between the presence of a mutation or the presence of many mutations that work together in response to a particular treatment, and the extreme effect of this is the response to immunotherapy. Immunotherapy checkpoint inhibitors work well in a small subset of patients—just around 20% in solid tumors—and there is a major effort in the oncology community to identify which patients will be likely responders to immunotherapy.

As we sequence the genomes of more patients, it is turning out to be the case that we can identify biomarkers of response to immunotherapy. What is increasingly obvious is that it is not one simple biomarker or one mutated gene that increases the likely response to immunotherapy. Rather, it is a combination of genes and a combination of mutations—sometimes mutations and genes that are not particularly important in terms of driving the cancer, but mutations that give rise to the ability of the immune cells and the immune system to recognize that certain cells are cancer cells, especially short mutated peptides that are exposed on the surface of the tumor itself.

The research we are performing also shows it is not only the number of mutations present in a tumor that is important to immunotherapy response, but what kinds of genes are expressed within the tumor and what they tell you about the kinds of immune cells that exist within that tumor. It is what we call gene-expression deconvolution.

When you sequence a tumor, you are also measuring the expression of genes coming from the immune cells, not just the tumor, and by looking at the expression signature of specific immune cells, you can uncover what kinds of immune cell populations you have within a tumor and how diverse immune cells are. We are finding that combining this type of immune characterization with tumor mutations and exposed peptides is predictive of who is likely to respond to immunotherapy.

Predicting Immunotherapy Response

How are you using this information to determine which patients are most likely to respond to -immunotherapy?

This is a very active area of research for us. We are combining information from all the different tumor signals we have been discussing—the number and type of mutations within a tumor and the number and type of mutations within the tumor’s immune cells—as well as measuring the number and diversity of T cells within a tumor microenvironment, which is also useful in predicting immunotherapy response.

This is an ambitious vision, but one that could result in ensuring patients only get a therapy that is very likely to work.

— Olivier Elemento, PhD

Tweet this quote

We are combining all this different genomic information into a machine learning–based predictive model we call the immune response index. The immune response index seeks to quantify with greater certainty the likely clinical outcome of immunotherapy. We are in a very early stage of research with this technology, but preliminary results show we are able to predict with between 85% and 95% accuracy which patients will respond to some forms of immunotherapy.

We need to independently validate our findings in larger cohorts of patients from other cancer centers, but our research indicates that by combining all types of genomic information, including tumor mutations, tumor RNA expression, and T cell–receptor gene expression, we are able to make very accurate predictions on immunotherapy response.

The results from our study are under review, and we hope to see them published soon.

Integrating Whole-Genome Sequencing Into Routine Clinical Care

When will you transition to routinely using next-generation whole-genome sequencing on your patients with cancer?

I think within the next 3 to 4 years, we will systematically be doing whole-genome sequencing on the majority of our patients. But just determining the complete DNA sequence of a tumor’s genome and identifying large numbers of relevant mutations and copy number alterations are not enough to give us a full picture of what is actually driving the cancer, especially once the cancer metastasizes to other organ sites.

We know, for example, that tumors evolve as they move from site to site, and the driver mutations that might be important to discern for treatment at the primary site might not be as important once the tumor changes locations, because the environment and associated growth signals—for example, hormones—are different. So, it is critically important for us to be able to identify the driver mutations in both the primary and metastasized tumors.

In addition to whole-genome sequencing, it is now clear that genomic testing platforms need to employ a multiomic approach combining DNA sequencing with a more functional readout of the genome. For example, transcriptome sequencing tells us the expression status of the mutated genes and reveals whether these mutations truly drive tumors or have simply been inherited from a previous tumor stage and have ceased to be drivers. We can extend this analysis of genes downstream of specific mutations, asking whether the pathway these mutations fall in is aberrantly active or suppressed. Such mutations are likely to be true drivers of the disease.

We are also researching a “disease in a Petri dish” platform, which uses three-dimensional models of diseased cells grown from patients’ tumor cells, called tumor organoids, to make a lot of mini tumors out of the original tumor and then perform high-throughput drug screening. It’s like running a mini clinical trial with thousands of drugs and drug combinations for individual patients. After identifying a potential treatment, the organoid is transplanted into a mouse model to assess how well the tumor responds to the drug. The organoid cell culture system drastically improves the identification of potential therapies compared with a genomics-only approach. It sometimes identifies effective therapies in cases where no well-known driver mutations have been found.

The idea surrounding tumor organoids is the same as the one for the functional readout of the genome: to systematically test hypotheses to learn how likely it is that a patient will respond to a specific drug, without testing the drug on the patient first.

Ushering in the Era of Precision Medicine

Would this integration of whole-exome sequencing with a living biobank that enables high-throughput drug screening on patient-derived tumor organoids allow oncologists to truly provide personalized medicine for every patient with cancer?

Yes. This is an ambitious vision, but one that could result in ensuring patients only get a therapy that is very likely to work.

In a study we recently published, we established 56 tumor-derived organoid cultures and 19 patient-derived xenograft models from 769

Patient-derived organoid technology holds great promise for disease modeling, toxicity testing, discovery of novel therapies, and truly guiding precision medicine in oncology care.

— Olivier Elemento, PhD

Tweet this quote

patients with advanced cancer enrolled in a clinical trial. After using high-throughput drug screening to learn effective treatment strategies, we analyzed tumor-derived cells from four patients and were able to identify effective drugs and drug combinations in two cases of uterine cancer and two cases of colorectal cancer.²

Our greatest challenge, however, is that we are not able to grow many patient-derived organoid cells in a Petri dish. And organoids have shortcomings. They are devoid of stromal cells, blood vessels, and immune cells, and it is not clear whether organoids can capture the heterogeneous gene cell populations stemming from the original tumor. So we need to learn how to make better organoids—that is, organoids more like the tumors seen in patients. For example, we need to learn about how to develop co-culture conditions of organoids with immune cells or other cells, including with such cells coming from the same patient as the tumor cells. This is an active area of research at the Englander Institute for Precision Medicine.

Still, patient-derived organoid technology holds great promise for disease modeling, toxicity testing, discovery of novel therapies, and truly guiding precision medicine in oncology care. ■

DISCLOSURE: Dr. Elemento is Director of the Englander Institute for Precision Medicine, where the Exome Cancer Test was developed.

REFERENCES

1. Rennert H, Eng K, et al: Development and validation of a whole-exome sequencing test for simultaneous detection of point mutations, indels and copy-number alterations for precision cancer care. NPJ Genom Med. July 20, 2016 (Epub).

2. Pauli C, Hopkins BD, Prandi D, et al: Personalized in vitro and in vivo cancer models to guide precision medicine. Cancer Discov 7:462-477, 2017.